Effects of thoracic aortic occlusion and cerebrospinal fluid drainage on regional spinal cord blood flow in dogs: Correlation with neurologic outcome

Transcription

1 Effects of thoracic aortic occlusion and cerebrospinal fluid drainage on regional spinal cord blood flow in dogs: Correlation with neurologic outcome Thomas C. Bower, MD, Michael J. Murray, MD, Phi), Peter Gloviczki, MD, Tony L. Yaksh, PhD, Larry H. Hollier, MD, and Peter C. Pairolero, MD, Rochester, Minn. We studied the effect of thoracic aortic occlusion and cerebrospinal fluid (CSF) drainage on regional spinal cord blood flow and its correlation with neurologic outcome. Using isotope-tagged microspheres, we determined blood flow to the gray and white matter of five regions of the spinal cord in dogs: group I (control), group II (cross-damp only), group III (cross-clamp plus CSF drainage). At 6 minutes after thoracic aortic occlusion in group II, median gray matter blood flow (GMBF) in the lower thoracic and lumbar cord decreased from 23.1 and 27. ml/1 gm/min at baseline to 4. and 2.5 ml/1 gm/min, respectively. The addition of CSF drainage improved GMBF during aortic crossdamping in the lower thoracic and lumbar cord to 11.3 (p <.5) and 15.1 ml/1 gm/min (p <.3), respectively. After removal of the aortic cross-clamp, median blood flow more than tripled from baseline blood flow in group II, whereas CSF drainage prevented significant reperfusion hyperemia. Both low GMBF during cross-clamping and reperfusion hyperemia were associated with a worse neurologic outcome. In group II, no dog was neurologically normal, and more than 6% of the dogs had spastic paraplegia. In contrast, almost 6% of dogs in group HI were normal, and none had spastic paraplegia (p <.1). We conclude that CSF drainage in dogs during thoracic aortic occlusion maintained spinal cord perfusion above critical levels, diminished reperfusion hyperemia, and improved neurologic outcome. (J VAsc SUgG 1988;9: ) Paraplegia continues to be one of the most devastating complications after reconstructions of the descending thoracic and thoracoabdominal aorta. 1-s The incidence of paraplegia in different series 1-s varies from.4% to 4% depending on factors such as the type and extent of reconstruction, the presence of dissection, the duration of aortic cross-clamp time, and the age of the patient. Numerous adjunctive techniques have been used clinically to prevent spinal cord ischemia, such as aortic shunts, 6,7 reimplantation of intercostal arteries, 2,3,s expeditious operation to minimize aortic cross-clamp time, hypothermia, 4 and cerebrospinal fluid (CSF) drainage alone 8 or with From the Section of Vascular Surgery, the Division of Intensive Care and Respiratory Therapy and the Neurologic Surgical Research Laboratory, Mayo Clinic and Mayo Foundation. Presented at the Forty-second Annual Meeting of the Society of Vascular Surgery, Chicago, Ill., June 13-15, Reprint requests: Peter Gloviczld, MD, Section of Vascular Surgery, Mayo Clinic, 2 First St. SW, Rochester, MN intrathecal administration of papaverine. 9 Although several of these techniques appear beneficial, none has consistently prevented paraplegia in large numbers of patients undergoing thoracoabdominal reconstruction. Recent experimental studies from our laboratory confirmed the effectiveness of CSF drainage in improving arterial perfusion pressure in the spinal cord and decreasing the incidence of neurologic damage in dogs after cross-clamping of the thoracic aorta. 8 In baboons spinal fluid drainage with the addition of papaverine also prevented the development of paraplegia. 9'1 In an effort to understand further the mechanisms involved in producing paraplegia, our present study focused on determining the temporal changes in regional spinal cord blood flow in dogs during thoracic aortic cross-clamping, with or without CSF drainage. In addition, correlations between changes in segmental spinal cord blood flow, specifically to the gray and white matter, and clinical outcome were determined. 135

2 136 B~etM. Journal of VASCULAR SURGERY Carotid MAP A Femoral MAP B ] PA Pressure I CSF Pressure Fig. 1. Canine model used in this experiment. MATERIAL AND METHODS Twenty-one mongrel dogs (weighing 19 to 26 kg) were classified into three groups. Group I consisted of six control dogs that had a sham left thoracotomy without aortic cross-clamping. Group II consisted of eight dogs that had descending thoracic aortic occlusion with a single cross-clamp for 6 minutes without CSF drainage. Group III comprised seven dogs that had CSF drainage followed by descending thoracic aortic occlusion for 6 minutes. Spinal cord blood flow was measured in each group by isotope-tagged microspheres before, during, and after aortic occlusion. Experimental design. Anesthesia was induced with intravenous methohexital (1 mg/kg). The dogs were intubated and ventilation was maintained with a blend of room air and inhalation anesthesia (halothane, 1% to 1.5%) delivered by a Bird Mark 7 respirator. Succinylcholine (2 mg) was administered to provide skeletal muscle relaxation. Approximately 1 ml of non-dextrosecontaining Plasmalyte was administered intravenously during the course of the procedure. A cerebrospinal catheter to monitor CSF pressure and to withdraw CSF was introduced via the cisterna magna under direct vision to avoid CSF leakage. Left carotid and femoral arterial lines were placed to monitor blood pressure above and below the level of the aortic cross-clamp and to sample arterial blood gases. A Swan-Ganz catheter was inserted to monitor core temperature and pulmonary arterial pressure and to calculate cardiac output by the thermodilution method. All pressure lines were connected to strain- gauge transducers and pressures were recorded continuously throughout the experiment on a strip chart recorder (Fig. 1). A warming blanket and heating lamp were used to maintain a constant core temperature. A left thoracotomy was pcrformcd through thc fifth intercostal spacc. Alcft atrial line was placcd to inject the microspheres into the central circulation. The descending thoracic aorta was exposed approximately 1 cm distal to the origin of the left subclavian artery but above the origin of the highest aortic intercostal arteries. In group III animals, CSF was drained until no further fluid could be aspirated. After intravenous injection of heparin (1 U/kg), the aorta was cross-clamped 1 cm below the left subclavian artery for 6 minutes in the animals in groups II and III. Sodium bicarbonate (2 to 25 meq) was administered intravenously before the aortic cross-clamp was removed to minimize the effects of metabolic acidosis. At the end of the procedure the thoracotomy was closed, air was aspirated from the pleural space, and the animal was monitored for 24 hours to assess neurologic outcome. At 24 hours the animals were killed with an overdose of intravenous pentobarbital. Animal care complied with the "Principles of Laboratory Animal Care" and the "Guide for the Care and Use of Laboratory Animals," (NIH Publication No. 8-23, revised 1985), and this study was approved by our Institutional Animal Care Committee. Microsphere measurement technique. Spinal cord blood flow was determined by the use of one

3 Volume 9 Number 1 Janua~ 1989 Regional spinal cord blood flow in dogs 137 Proximal mean aortic pressure, mm Hg 12 I 1 F 8/ II I I I 12 o o Distal mean aortic pressure, mm Hg 6 15 II I I I Cerebrospinal fluid pressure, mm Hg 1 5 3O Spinal cord perfusion pressure, mm Hg 2-1- [3. /r [3 II Cross-clamp on I I I 5 min 6 rain 65 min Cross-clamp off Fig. 2. Hemodynamic measurements before, during, and after thoracic aortic occlusion. o o, Group I (control);, group II (cross-clamp); [] [], group III (cross-clamp plus CSF drainage). offourmicrospheres (15 +_ 3 ~m),labeledwithszco (cobalt), 113Sn (tin), SSSr (strontium), or 46Sc (scandium) (New England Nuclear, Billerica, Mass.). Four different tagged microspheres were used to prevent ordering bias. The microspheres were suspended in a well-mixed solution of 1% dextran diluted to 3 ml with isotonic saline solution. Blood flow measurements were performed before aortic crossclamping, at 5 and 6 minutes during aortic crossdamping, and 5 minutes after the release of the crossdamp. For each blood flow measurement, reference samples were withdrawn from the carotid arterial line by an infusion-withdrawal pump at 7.41 ml/min for 9 seconds. Fifteen seconds after the withdrawal was started, the agitated microspheres (approximately ) were injected into the left atrium. Behavioral assessment. Neurologic assessment was noted at 24 hours and graded according to the method of TarlovU: --spastic paraplegia, no movement of the joints of the lower limbs; 1 = spastic paraplegia, perceptible movement of the joints of the lower limbs; 2 = good movement of the joints but unable to stand; 3 = able to stand but unable to walk normally; 4 = complete recovery. Calculation of blood flow. Immediately after the animals were killed, the spinal cord was removed from the foramen magnum to the sacrum, placed in buffered formalin for 48 hours, and divided into five regions: cervical, upper thoracic, midthoracic, lower thoracic, and lumbar. Each region was subdivided by sharp dissection into gray and white matter and weighed on a Mettler P121 scale to the nearest.1 gm. Samples were counted in a well-type gamma counter, with a Canberra series 35 plus multichannel analyzer to visually set the appropriate windows for the energy spectra of each microsphere. The number

4 138 Bower et al. Journal of VASCULAR SURGERY l Cervical I l~ uppe~ Thoracic I mid lower ~ l Lumbar I c- E O3 4 I [] Gray matter 3 [] White matter 2 -- " m 1 -- Fig. 3. Baseline gray and white matter blood flow (median) in various regions of spinal cord. of counts from the specific microsphere was determined after correction of background and spillover. Blood flow was calculated by the methods outlined by Heymann et al. 12 Statistical methods. Comparisons of blood flow, neurologic outcome, and spinal cord perfusion among groups of dogs were performed with twotailed rank-sum tests. Correlations of blood flow with neurologic outcome in animals undergoing thoracic aortic occlusion (groups II and III) were performed with two-tailed tests based on Spearman rank correlation coefficients. Significance was defined as a p value ~<.5. RESULTS Hemodynamic measurements. In group I (control) animals, no difference was found among mean proximal aortic blood pressure, mean distal aortic blood pressure (BP), mean pulmonary artery pressure, or mean cardiac output at the four specific times analyzed (Fig. 2). No major differences were noted in hemodynamic values in group II compared with group III at the various times. Overall, core temperature ranged from 35.5 to 37.8 C. Less than.5 C variation was observed during the course of the experiment for each animal. Blood pressure. In group II animals (aortic cross-clamp, no CSF drainage), the mean proximal aortic BP increased from baseline values of to mm Hg at 6 minutes of aortic crossclamping. Simultaneously, the mean distal aortic BP decreased from an initial level of 81 _ 18 to 23 _ 7 mm Hg. Within 5 minutes after release of the aortic cross-clamp, mean proximal and distal BP returned to levels of 9 _+ 12 and mm Hg, respectively (Fig. 2). The mean proximal and distal aortic BP in group III animals (aortic cross-clamp, with CSF drainage) followed similar patterns. The mean proximal BP increased from mm Hg at baseline to 136 _ 15 mm Hg at 6 minutes of aortic crossdamping. Sharp decreases in mean distal BP occurred from mm Hg initially to mm Hg at 6 minutes of damping. Pulmonary artery pressure. The mean pulmonary arterial diastolic pressure (PADP) and mean pulmonary arterial pressure (PAP) increased in the animals in both groups (II and III) during the 6-minute aortic occlusion. In group II animals, PADP increased from 8.5 _+ 2.6 mm Hg at baseline to mm Hg by 6 minutes of crossclamping, and PAP increased from mm Hg initially to 16.2 _+ 5.1 mm Hg during the same period. In group III animals, PADP and PAP increased in a similar fashion. After release of the aortic cross-clamp, PADP and PAP returned to almost normal levels in the animals in groups II and III. Cardiac output. Baseline mean cardiac output was similar in group II (4. _ L/min) and group III dogs (3.78 _ L/rain). During aortic occlusion, cardiac output decreased in group II animals, reaching the lowest mean value of 2.3 _+.95 L/min at 6 minutes. This represented a decrease of 36% from initial baseline measurements. The mean cardiac output in group III animals also decreased during cross-clamping to L/min. After release of the aortic cross-clamp, the mean cardiac output was identical in both groups at 4.83 L/rain. Cerebrospinal fluid pressure. The mean baseline CSF pressure in group II dogs was 5.6 _+ 2.3

5 Volume 9 Number 1 January Regional spinal cord blood flow in dogs 139 Table I. Neurologic outcome (Tarlov score) with or without cerebrospinal drainage and thoracic aortic occlusion Tarlov score Group II (No.) Group III (No.) (paraplegia) (normal) 4 Group II, Cross-clamp, no CSF drainage; group III, cross-clamp plus CSF drainage. mm Hg. During aortic occlusion, mean CSF pressure increased to a maximum of 1.4 _ 3.8 mm Hg, an 85% increase from initial CSF pressure measurements. After release of the aortic cross-clamp, mean CSF pressure decreased to mm Hg. In group III animals, a mean of ml of CSF was removed, resulting in a decrease of mean CSF pressure from to mm Hg. Although CSF pressure increased during the aortic occlusion period, reaching a maximum level of 3. _+ 6.7 mm Hg at 6 minutes, it was three times less than the maximal CSF pressure recorded in group II animals at the same time. After aortic the clamp was removed, mean CSF pressure in group III animals decreased to mm Hg (Fig. 2). Spinal cord perfusion pressure. Spinal cord perfusion pressure (SCPP = mean distal aortic pressure minus mean CSF pressure) was calculated in the animals of groups II and III. During the period of aortic occlusion, SCPP decreased to 8.3 mm Hg in group II animals at 5 minutes and to 13.5 mm Hg at 6 minutes (Fig. 2). In contrast, in group III animals, SCPP at 5 and 6 minutes of aortic occlusion was 18. mm Hg (p <.1) and 21.1 mm Hg (p <.2), respectively. Arterial blood gas. Arterial ph and oxygen measurements were similar among groups, but Paco2 levels were lower in control dogs (range 26 to 28.5 torr) compared with levels of dogs in either group II (range 3 to 34 torr) or group III (range 26 to 41 torr). Neurologic outcome. At 24 hours all group I (control) animals were neurologically normal (Tarlov grade 4). Five of eight dogs in group II had spastic paraplegia of the hind limbs (Tarlov grade ). Two of the remaining three dogs showed good movement of the hind limbs but were unable to stand or walk (Tarlov grade 2). The last dog could stand but was r- 14 E _ O') 12 1 ~ 6 "(3 o.q 4 c- ~ 2 O---O Group I (control) -" -" Group II (cross-clamp) _ - - Group III (cross-clamp + CSF drainage) [ + 5 rain 6 rain + 65 rain Gross-clamp Cross-clamp on off Fig. 4. Temporal changes of gray matter blood flow in lumbar segment of spinal cord. extremely ataxic when attempting to walk (Tarlov grade 3). None of the dogs in this group was neurologically normal according to clinical observations. In contrast, animals with CSF drainage (group III) showed a significant difference in neurologic outcome (p <.1). Four of seven dogs were normal (Tarlov grade 4); one of three remaining dogs could stand and walk with minimal ataxia (Tarlov grade 3); the last two dogs showed varying degrees of paraparesis (Tarlov grades 1 and 2). No dog in this group had spastic paraplegia at 24 hours after aortic occlusion (Table I). Spinal cord blood flow. Median GMBF to each region of the spinal cord before cross-clamping was approximately two to three times greater than white matter blood flow (Fig. 3). This relationship was true for all further blood flow measurements in gray and white matter. Median GMBF in the five spinal cord regions varied little with time (Table II). Given the importance of GMBF for spinal motor neuron function, only gray matter flow will be subsequently emphasized. In group II animals, no difference was found in median cervical or upper thoracic GMBF during and after thoracic aortic occlusion compared with group I (control). Median midthoracic GMBF significantly decreased at 5 minutes of clamp time (p <.4) compared with blood flow in control animals. Blood flow to the lower thoracic and lumbar gray matter was significantly decreased at 5 minutes (p <.1 for both regions) and 6 minutes after aortic crossclamping (p <.1 andp <.1, respectively). After release of the aortic cross-clamp, the midthoracic, lower thoracic, and lumbar regions (Figs. 4 and 5) showed a significant hyperemic response compared

6 14 Bower et al. Journal of VASCULAR SURGERY Table II. Median gray matter blood flow in different regions of the spinal cord before, during, and after thoracic aortic occlusion Gray matter blood flow (ml/1 O gm/ rain) Region rain 5 rain 6 rain 65 rain Group I (control) Cervical 14.9 ( ) 18.2 ( ) 14.7 ( ) Upper thoracic 25. ( ) 26.1 ( ) 22.6 ( ) Midthoracic 22.1 ( ) 2.3 ( ) 2. ( ) Lower thoracic 25.5 ( ) 21.3 ( ) 17.4 ( ) Lumbar 26.8 ( ) 28.8 ( ) 24.1 ( ) Group II (cross-clamp) Cervical 21.2 ( ) 2.9 ( ) 27.6 ( ) Upper thoracic 28.2 (1-7.4) 16. ( ) 14. ( ) Midthoracic 21.5 (1-68.1) 8.4 ( ) 11.2 (4.3-2.) Lower thoracic 23.1 ( ) 2.5 ( ) 4. (-12.6) Lumbar 27. ( ) 1.4 ( ) 2.5 (.4-1.5) Group III (cross-clamp + (CSF drainage) Cervical 17.1 ( ) 19.4 ( ) 33.9 ( ) Upper thoracic 2.4 ( ) 11.9 (4-65.8) 16.1 ( ) Midthoracic 17.1 ( ) 1.1 ( ) 9. ( ) Lower thoracic 19. ( ) 14.4 (5-42.3) 11.3 ( ) Lumbar 21.8 ( ) 9.7 ( ) 15.1 (.5-4.3) 14.5 ( ) 22.2 ( ) 17.5 ( ) 21.5 ( ) 27.5 ( ) 28.3 ( ) 33.2 ( ) 65.5 ( ) 83.3 ( ) ( ) 3.7 ( ) 28.8 ( ) 3.4 (1.-7.) 28. ( ) 34.1 ( ) Data in parentheses indicate range. For groups II and III, values at 5 and 6 minutes are with cross-clamp on; at 65 minutes, the crossclamp was removed. with those in control animals (p <.5, p <.1, p <.1, respectively). In dogs with CSF drainage (group III), with the exception of midthoracic GMBF at 5 minutes of cross-clamping (p <.5), no significant differences were noted between median GMBF of the various regions of the spinal cord before, during, or after aortic clamping compared with control dogs. Blood flow to the cervical, upper thoracic, and midthoracic gray matter showed less change during and after thoracic aortic occlusion compared with that in group II animals (p = NS). Median GMBF to the lower thoracic and lumbar segments during aortic crossdamping also decreased in group III animals, but this change was not as great as in group II animals (Figs. 4 and 5). The differences in flow to these regions between groups II and III during the occlusion time were significant (p <.1 for both regions at 5 minutes of damping; p <.5 and p <.3, respectively, at 6 minutes of clamping). There was no reperfusion hyperemia, as noted in group II dogs, in the midthoracic, lower thoracic, or lumbar regions. The differences in reperfusion flow (Figs. 4 and 5) between the two groups were apparent primarily in the lower thoracic and lumbar gray matter (p <.6 andp <.2, respectively). Spinal cord blood flow and neurologic outcome. When GMBF data from the various cord regions were analyzed for dogs in groups II and III, significant correlations were noted for the lower spi- nal cord segment between blood flow and clinical neurologic outcome (Tarlov score). No correlation was found between cervical GMBF and neurologic outcome. However, decreased blood flow to the upper thoracic gray matter at 5 and 6 minutes of occlusion was associated with a better neurologic result (p <.2, Rs = -.64 at 5 minutes; p <.6, Rs = -.51 at 6 minutes). The hyperperfusion observed in this region and in the midthoracic gray matter after release of the cross-clamp in these animals was correlated with a poor neurologic outcome (p <.3, Rs = -.59; p <.1, Ks = -.75, respectively). In the lower thoracic gray matter during cross-clamping, low perfusion noted at both 5 and 6 minutes of aortic occlusion correlated with a lower Tarlov score (p <.1, Rs =.7, for 5 minutes; p <.1, Rs =.66, for 6 minutes). Low levels of blood flow to the lumbar region were related to worsening degrees of neurologic injury as well (p <.1, Rs =.67, at 5 minutes; p <.1, Rs =.76, at 6 minutes). After release of the aortic cross-damp, the marked increase in lower thoracic GMBF was associated with a poor neurologic outcome (p <.3, Rs = -.56); however, this same relationship was not evident in the lumbar gray matter, although a similar increase in GMBF occurred. Significance may not have been reached for the lumbar gray matter because one dog with spastic paraplegia (Tarlov grade ) had a reperfusion blood flow of approximately 18 ml/1 grn/min, whereas the

7 Volume 9 Number 1 January Regional spinal cord blood flow in dogs 141 c_ E o Cerwcal 4. 3o T~oracic 41i' Lumbar i ' 3O i Cervical Thoracic rn~d lower ~ ~ ~,ota Lumber -~ 2o 2 E zo ~ 3o 8 2O t I I I L O~O Group I (controlt Group II (cross-clamp/ H Group III (cross-clamp ~1 I I I T ainage) I I I I I Fig. 5. Regional gray matter blood flow (median) in spinal cord before, during, and after thoracic aortic occlusion. A, Before aortic cross-clamping. B, Five minutes after aortic crossclamping. C, Sixty minutes after aortic cross-clamping. D, Five minutes after aortic clamp removal. remaining dogs with rigid paraplegia demonstrated =_ 45 much higher flows, approximately 124 to 18 ~ 4 ml/1 gm/min g 3s Individual animals in groups II and III that be- ~ 3 came markedly paraparetic (Tarlov grade 1) or parao ~ 2s plegic (Tarlov grade ) showed a tendency for re- _o 2 duced blood flow in the lumbar cord during thoracic ~s aortic occlusion (Fig. 6) and increased flow during -~ r- 1 reperfusion. Conversely, those animals with corn-._~ "o S plete recovery (Tarlov grade 4) or minimal paraparesis (Tarlov grade 3) showed a tendency toward more normal (baseline) flow during the occlusion with little hyperemic response after release of the cross-clamp. Similar changes were seen with respect to lower thoracic GMBF in these animals. 4. Paraplegic Group II (cross-clamp) Group III (cross-clamp + CSF drainage) BE I I I Normal Tarlov score Fig. 6. Correlation of lumbar cord blood flow at 6 minutes of aortic cross-clamping and clinical outcome. DISCUSSION In accordance with data obtained in baboons by Svensson et al., * we have shown in dogs that regional spinal cord blood flow can be measured with the microsphere technique and that these measurements are reproducible under control circumstances, provided that cardiac output, arterial blood pressure, and arterial blood gases remain within the limits of autoregulation for the spinal cord? sq6 The overall baseline flow rates in our study are comparable to those seen in other studies. 1'16-19 Although experimental evidence has shown that reductions in aortic pressure below the level of cross- clamping change the amount of blood distributed to the distal spinal cord, 17 few data have been generated regarding regional ~,17'*s or, more specifically, gray matter spinal cord blood flow, which would affect spinal motor neurons. Our study showed that regional changes in spinal cord blood flow occurred during and after thoracic aortic occlusion, with the most dramatic changes in blood flow occurring in those regions located the farthest from the level of the cross-clamp, that is, the lower thoracic and lumbar cord. In addition, using an experimental model and occlusion time that have produced high rates of

8 142 Bower et a/. Journal of VASCULAR SURGERY paraplegia, 9,2 '21 we have shown that CSF drainage not only decreased the incidence of neurologic injury but also resulted in significant improvement in spinal cord blood flow compared with animals without CSF drainage. Two important issues are apparent from our data. CSF drainage before aortic occlusion allowed adequate spinal cord perfusion, which significantly decreased the incidence of neurologic complications otherwise observed during thoracic aortic crossclamping. The marked hyperemic response during reperfusion seen in animals undergoing crossclamping only was not observed in those animals receiving CSF drainage. Some studies 1 '22 have been unable to confirm significant improvement in neurologic outcome with CSF drainage alone. Svensson et al. 1 failed to observe neurologic benefit from CSF drainage alone via laminectomy in a study in baboons, although blood flows to the lower thoracic and lumbar regions of the spinal cord at the end of 6 minutes of crossclamping were improved compared with that in control animals. They did find that the administration of intrathecal papaverine in addition to drainage of CSF, or the use of aortic shunts during crossclamping with occlusion of the infrarenal aorta, entirely prevented paraplegia. However, other investigators found similar decreased rates of paraplegia when CSF was drained and the aorta was cross-clamped for 6 minutes. 9'2 '21 Paraplegia occurred in 5% of surviving dogs in the study of Blaisdell and Cooley 21 vs 8% in dogs with CSF pressure decompression. All dogs in that study with CSF pressures greater than or equal to distal aortic pressures became paraplegic. Miyamoto et al.2 showed similar benefits with CSF drainage; paraplegia rates decreased from 75% in dogs without CSF drainage to 5% with CSF drainage. McCullough et al. 8 from our laboratory demonstrated significant reductions in paraplegia or paraparesis in dogs with CSF drainage. These investigators postulated that in addition to improving the effective SCPP, CSF drainage may allow decompression of high-resistance spinal collateral arteries. The latter phenomenon may be occurring because paraplegia has been reported both with low and high distal aortic perfusion pressures, 2'a's'7 as well as with increased CSF pressure. 23 Although neither the minimal blood flow nor the duration of ischemia tolerated by the spinal cord is known, certain areas of the brain subjected to marginal flow may exist in a nonfunctional state that is reversible Additional small increases in blood flow, once this reversible brain ischemia situation is reached, may result in rapid recovery, or total loss of function if perfusion continues to decrease, z4-27 If nerve cells in the spinal cord react to ischemia as do nerve cells in the brain, then maintaining flow to regions of the spinal cord rendered ischemic during cross-clamping becomes important. Even at relatively low distal aortic pressures, small changes in flow, such as with CSF drainage, may allow enough improvement in perfusion to alter neurologic ischemic injury. 27,2s The second important issue suggested by our findings is that CSF drainage is associated with a marked decrease in local hyperemia otherwise observed after declamping. For example, reperfusion after ischemia to brain tissue may result in injury caused by changes in the blood-brain barrier or in decreased perfusion because of elevated cerebral vascular volume. 2s'29'3 Whether reperfusion causes the actual damage, contributes to the sequelae, or is only a response to the injury is not dear. The magnitude of the hyperemic reflow response does appear to be an index of the degree to which the integrity of the vascular autoregulatory mechanisms has been affected by the ischemic insult. These observations support the contention that reduction of CSF pressure during low flow conditions seems to be a protective measure in maintaining the physiologic function of the spinal vascular tree. In conclusion, we have shown that blood flow to the distal spinal cord decreases significantly during thoracic aortic occlusion in dogs. The addition of CSF drainage improves gray matter blood flow to the lower thoracic and lumbar cord during aortic occlusion, prevents reperfusion hyperemia, and improves neurologic outcome. These findings justify further investigations of CSF drainage in the clinical setting. REFERENCES 1. Brewer LA IlL Fosburg RG~ Mulder GA, Verska JJ. Spinal cord complications following surgery for coarctation of the aorta: a study of 66 cases. J Thorac Cardiovasc Surg 1972;64: Crawford ES, Rubio PA. Reappraisal of adjuncts to avoid ischemia in the treatment ofaneurysms of descending thoracic aorta. J Thorac Cardiovasc Surg 1973~66: Livesay JJ, Cooley DA, Ventemiglia RA, et al. Surgical experience in descending thoracic aneurysmectomy with and without adjuncts to avoid ischemia. Ann Thorac Surg 1985;39: Laschinger JC, Izumoto H, Kouchoukos NT. Evolving concepts in prevention of spinal cord injury during operations on the descending thoracic and thoracoabdominal aorta. Ann Thorac Surg 1987;44:

9 Volume 9 Number 1 January Regional spinal cord blood flow in dogs Crawford ES, Crawford JL, Sail HJ, et al. Thoracoabdominal aortic aneurysms: preoperative and intraoperative factors determining immediate and long-term results of operations in 65 patients. J VAsc SUV, G 1986;3: Carlson DE, Karp RB, Kouchoukos NT. Surgical treatment of aneurysms of the descending thoracic aorta: an analysis of 85 patients. Ann Thorac Surg 1983;35: Donahoo JS, Brawley RK, Gott VL. The heparin-coated vascular shunt for thoracic aortic and great vessel procedures: a ten-year experience. Ann Thorac Surg 1977;23: McCullough JL, HoUier LH, Nugent M. Paraplegia after thoracic aortic occlusion: influence of cerebrospinal fluid drainage; experimental and early clinical results. J VASC SUV, G 1988;7: Svensson LG, Loop FD. Prevention of spinal cord ischcmia in aortic surgery. In: Bergan JJ, Yao JST, cds. Arterial surgery: new diagnostic and operative techniques. Orlando: Grunt & Stratton, Inc, 1988: Sveusson LG, Von Ritter CM, Grocneveld HT, et al. Crossclamping of the thoracic aorta: influence of aortic shunts, laminectomy, papaverine, calcium channel blocker, allopurinol, and superoxide dismutase on spinal cord blood flow and paraplegia in baboons. Ann Surg 1986;24: Tarlov IM. Spinal cord compression: mechanism of paralysis and treatment. Springfield, Ill: Charles C Thomas, Heymann MA, Payne BD, Hoffman lie, Rudolph AM. Blood flow measurements with radionuclide-labeled particles. Prog Cardiovasc Dis 1977;2: Hickey R, Albin MS, Bunegin L, Gelineau J. Autoregulation of spinal cord blood flow: is the cord a microcosm of the brain? Stroke 1986;17: Griffiths IR. Spinal cord blood flow in dogs: the effect of blood pressure. J Neurol Neurosurg Psyehiatr 1973;36: Griffiths IR. Spinal cord blood flow in dogs. 2. The effect of the blood gases. J Neurol Neurosurg Psyehiatr 1973;36: Marcus ML, Heistad DD, Ehrhardt JC, Abboud FM. Reguiation of total and regional spinal cord blood flow. Circ Res 1977;41: Gelman S, Reves JG, Fowler K, Samuelson PN, Lell WA, Smith LR. Regional blood flow during cross-clamping of the thoracic aorta and infusion of sodium nitroprusside. J Thorac Cardiovasc Surg 1983;85: Svensson LG, Rickards E, Coull A, Fimmel CJ, Hinder RA. Relationship of spinal cord blood flow to vascular anatomy during thoracic aortic cross-clamping and shunting. J Thorac Cardiovasc Surg 1986;91: Grifliths IR, Rowan JO, Crawford RA. Spinal cord blood flow measured by a hydrogen clearance technique. J Neurol Sci 1975;26: Miyamoto K, Ueno A, Wada T, Kimoto S. A new and simple method of preventing spinal cord damage following temporary occlusion of the thoracic aorta by draining the cerebrospinal fluid. J Cardiovasc Surg 196; 1: Blaisdell FW, Cooley DA. The mechanism of paraplegia after temporary thoracic aortic occlusion and its relationship to spinal fluid pressure. Surgery 1962;51: Wadouh F, Lindemann EM, Amdt CF, Hetzer R, Borst HG. The arteria radicularis magna anterior as a decisive factor influencing spinal cord damage during aortic occlusion. J Thorac Cardiovasc Surg 1984;88: Griftiths IR, Pitts LH, Crawford RA, Trench JG. Spinal cord compression and blood flow. I. The effect of raised cerebrospinal fluid pressure on spinal cord blood flow. Neurology 1978;28:1, Meyer FB, Piepgras DG, Sandok BA, Sun& TM Jr, Forbes G. Emergency carotid endarterectomy for patients with acute carotid occlusion and profound neurological deficits. Ann Surg 1986;23: Sundt TM Jr, Sharbrough FW, Marsh WR, Ebersold MJ, Piepgras DG, Messick JM Jr. The risk-benefit ratio of intraoperative shunting during carotid endarterectomy: relevancy to operative and postoperative results and complications. Ann Surg 1986;23: Sundt TM Jr, Sharbrough FW, Piepgras DG, Kearns TP, Messick JM Jr, O'Fallon WM. Correlation of cerebral blood flow and electroencephalographic changes during carotid endarterectomy: with results of surgery and hemodynamics of cerebral ischemia. Mayo Clin Proc 1981;56: HoUier LH. Protecting the brain and spinal cord. J Vasc SURG 1987;5: Messick JM Jr, Newberg LA, Nugent M, Faust RJ. Principles of neuroanesthesia for the nonneurosurgical patient with CNS pathophysiology. Anesth Analg 1985;64: Sundt TM Jr, Waltz AG. Cerebral ischemia and reactive hyperemia: studies of cortical blood flow and microcirculation before, during, and after temporary occlusion of middle cerebral artery of squirrel monkeys. Circ Res 1971;28: Lassen NA. The luxury-perfusion syndrome and its possible relation to acute metabolic acidosis localised within the brain. Lancet 1966;2: DISCUSSION Dr. Ronald J. Stoney (San Francisco, Calif.). Dr. Bower and his coauthors are to be congratulated on a wellwritten article. This is an important topic that considers experimental spinal cord ischemia and cerebrospinal fluid pressure reduction, to modify the neurologic outcome of that insuk. This observation is not actually new. Blaisdell and Cooley, among others, showed a dramatic reduction in the paraplegic incidence in dogs by spinal fluid drainage. They also showed that if spinal fluid pressure exceeded or equalled the distal aortic pressure, paraplegia occurred universally. I think this is not surprising when you consider the dog has an incomplete anterior spinal artery and that the distal spinal cord perfusion is absolutely dependent on blood flow from the aorta and its branches distal to the damp. However, human dissection studies by Stinson and others have shown in primates, especially in man, a continuity of the anterior spinal artery in all these dissected studies. Therefore, it may explain the species difference in

10 144 Bower et al. Journal of VASCULAR SURGERY the neurologic outcome of humans and mongrel dogs subjected to simple thoracic aortic cross clamping. The neuroligic defects we have encountered from surgery in the thoraco-abdominal aorta or the supraceliac aorta have not been in the more than one hundred simple occlusions of the thoracic or supra celiac aorta when we did infrarenal aortic manipulations, but rather only in cases where there has been thoraco-abdominal aortic reconstruction of some type. Another source or mechanism that I believe is an important cause of cord ischemia in humans is embolization of the debris from the thoraco-abdominal aorta either during clamping, grafting, or other manipulations. The ostia of these critical low intercostals are obstructed or trashed with athero emboli. Dr. Bower, I wonder if you could speculate as to in what type of operation involving the thoraco-abdominal aorta would you consider the use of spinal cord fluid drainage? Is there a difference between patients having simple procedures receiving temporary thoraco-abdominal aortic clamping and those who undergo a more complex reconstruction? My final point of discussion that I think may be the most important point of this article concerns reperfiasion injury. The authors showed that there was a marked decrease in spinal cord hyperemia that normally follows prolonged ischemia in those dogs undergoing spinal cord fluid drainage. We all know about reperfusion injury, particularly in the brain where we are concerned about revascularizing the ischemic or pale infarct by carotid TEA after acute stroke. It occurs to me that this may be one of the mechanisms of spinal cord dysfimction with resulting paraplegia or paraparesis in humans. Is it the ischemic injury, or is it possibly that the reperfusion modifies and converts mild injury to a severe one like it has been known to do in the human brain? I would like to know what the authors think about the importance ofreperfiasion that they are able to modify with spinal cord fluid drainage? Could that be the most significant aspect of the neurologic injury? In other words, is it ischemia and do they have any idea about this, and possibly some ideas to separate these two factors and study the two in a similar experimental model? Dr. F. William Blaisdell (Sacramento, Calif.). Our study correlated the duration of occlusion and the spinal fluid pressure with the number of intercostal arteries divided. We found that the more intercostal arteries divided in our model, the greater the incidence of paraplegia, and this tends to emphasize that there are two factors involved in paraplegia. One is the ischemia generated by the crossclamp, the low perfusion pressure, and the high cerebrospinal fluid pressure: the second factor is the critical contribution of the division of intercostal arteries. Unfortunately, it is my impression that the greater risk is the division of critical intercostal arteries rather than this pressure differential. I believe the spinal cord tolerates approximately 3 minutes of ischemia. What it does not tolerate is interruption of critical intercostal arteries and I do not think we can provide any protection by modifying spinal fluid pressure. This is the primary reason that we have not used this technique over the years. Critical to this experiment is allowing the proximal pressure to increase. As the proximal pressure increases, the brain swells and that pressure is transmitted to the spinal cord. However, if one controls the proximal pressure by the appropriate use ofvasodilating drugs, then one can accomplish the same thing as spinal fluid drainage and improve the differential pressure. I wonder whether you have tried controlling the proximal pressure and its relationship to the incidence of paraplegia. Dr. Bower (Closing). In regard to Dr. Stoney's question, the exact role of cerebrospinal fluid drainage is yet to be determined but I do believe it can play a role in minimizing neurologic injury during repair of extensive thoracoabdominal aneurysms and aortic dissections. We have used this technique in 36 patients and have had no instances of paraplegia, but obviously these numbers are too small to make any conclusions at this time. Dr. Stoney has alluded to two important areas. The first concerns the applicability of the model, based on differences in spinal cord anatomy between the dog and the human being. Although other investigators have reported the canine anterior spinal artery to be discontinuous, the articles they cite actually have little information specifically concerning the dog's spinal vascular supply. With the use of selective spinal angiography, latex injections, and corrosion casting in 12 dogs, we have found that the anterior spinal artery of the dog is continuous, like the human. Therefore, with respect to the spinal cord vascular anatomy, we believe that the dog is certainly a good model for studying spinal ischemia. The second issue raised by Dr. Stoney concerns reperfusion injury. We are not sure when the neurologic injury occurs, whether it is during the ischemic period or in the reperfusion phase. Nevertheless, we believe that the marked hyperemia that occurs after release of the crossclamp indicates loss of the autoregulatory function of the spinal vascular tree. In our other ongoing studies we might be able to answer this question with this model. I certainly appreciate Dr. Blaisdell's comments, who was one of the first investigators to look at this technique. We have studied cerebrospinal fluid pressure changes experimentally during control of proximal arterial blood pressure. In several animals we have used nitroprusside and arbitrarily returned systolic blood pressure to levels before cross-clamping. This actually increased cerebrospinal fluid pressure and decreased the mean distal aortic pressure, thereby diminishing spinal cord perfusion. However, I do believe there will be a role for the use ofvasodilating agents, either selectively or systemically.